Egor Kuklin

I would like to take a shot at this If its not too early to try to adopt WebGPU that is. (I've read somewhere that its not finalized and might...

@gkjohnson Let me know if you think there should be a better solution

As I understand, current intersection algorithm is based on Moller's paper. I've done a bit of reading and there seems to be a more robust [algorithm](https://inria.hal.science/inria-00072100/document). However, in the book...

I've started working on explicitly handling degenerate cases before your comment and pushed results here if you want to check out how this might look anyway. I'm not particularly fond...

I have implemented Moller's algorithm from [here](https://web.stanford.edu/class/cs277/resources/papers/Moller1997b.pdf). It handles corner cases for intersections pretty well, but still with quite a number of ifs. Also, I've found a couple optimizations that...

> Can you explain this a bit more? The SAT axes code works by projecting points from both onto a given axis and checking of the min / max of...

Hi, yeah you can take a look at the new random benchmarks if you'd like. I've tried to stabilize the bench results but some margin for error still remains. For...

Hi @gkjohnson I think I've done everything I wanted to here, so you can take a look. Updated PR description to reflect everything done here. Explicit handling takes a lot...

> Have you already run them relative to the previous implementation? Can you share the performance difference? Yes, you can find them in the PR description. > I've also left...

Thank you for the feedback! > the performance seems at least on-par with the current one when integrating with bvh-csg Great! Let me know if bvh-csg has issues with the...